Research map / Image generation / Editing and style transfer
Follow-Your-Shape: Shape-Aware Image Editing via Trajectory-Guided Region Control
Zeqian Long, Mingzhe Zheng, Kunyu Feng, Xinhua Zhang, Hongyu Liu, Harry Yang, Linfeng Zhang, Qifeng Chen, Yue Ma
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use k_front=2 initial KV-injection steps instead of 0 (ablation) | Follow-Your-Shape · ReShapeBench | PSNR: 32.79 -> 35.79 | Improved |
| use k_front=2 instead of 0 (ablation) | Follow-Your-Shape · ReShapeBench | CLIP similarity: 31.05 -> 33.71 | Improved |
| increase k_front from 2 to 4 (ablation) | Follow-Your-Shape · ReShapeBench | PSNR: 35.79 -> 30.41 | Hurt |
| increase k_front from 2 to 4 (ablation) | Follow-Your-Shape · ReShapeBench | CLIP similarity: 33.71 -> 27.66 | Hurt |
| add ControlNet (depth and Canny) conditioning (ablation) | Follow-Your-Shape · ReShapeBench | LPIPS x10^3: 9.04 -> 8.23 | Improved |
| add ControlNet conditioning (ablation) | Follow-Your-Shape · PIE-Bench | PSNR: 35.62 -> 36.02 | Improved |
| remove ControlNet (ablation) | Follow-Your-Shape · ReShapeBench | aesthetic score: 6.57 -> 6.52 | No clear effect |
| inject ControlNet earlier in the denoising interval (ablation) | Follow-Your-Shape · ReShapeBench | qualitative | Improved |
| use overly weak or strong ControlNet strengths vs (2.5, 3.5) (ablation) | Follow-Your-Shape · ReShapeBench | qualitative | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- Dit4Edit — DiT4Edit: Diffusion Transformer for Image Editing (AAAI 2025) · better on 1 of 1 shared result
- FlowEdit — FlowEdit: Inversion-Free Text-Based Editing Using Pre-Trained Flow Models (ICCV 2025) · better on 1 of 1 shared result
- KV-Edit — KV-Edit: Training-Free Image Editing for Precise Background Preservation (ICCV 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Shape-Aware Text-Driven Layered Video Editing (CVPR 2023 · similarity 0.88)
- Blended Point Cloud Diffusion for Localized Text-guided Shape Editing (ICCV 2025 · similarity 0.88)
- GeoEdit: Geometry-Aware Object Editing via Dual-Branch Denoising (ECCV 2026 · similarity 0.87)
- CTRL-D: Controllable Dynamic 3D Scene Editing with Personalized 2D Diffusion (CVPR 2025 · similarity 0.86)
- Object-Aware Inversion and Reassembly for Image Editing (ICLR 2024 · similarity 0.86)
- PrEditor3D: Fast and Precise 3D Shape Editing (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
